Mar 02rd, 2016 at 5:00pm PDT, Agenda and Minutes
Don Brutzman, William Glascoe, Myeong W. Lee, Joe Williams, Dick Puk
Questions in the interface Humanoid:
1) Why skinNormal is necessary?
I think skinNormal is an animation option that might be useful for rendering effects.
Current graphics cards can rapidly compute normals so there is potential size/transmission performance penalty if they are used.
I see no need to remove that capability. There are probably some good character examples out there (future HAnim models!) that use normal animation.
2) Why bboxCenter and bboxSize are necessary?
Again optional. In this case, they can be very useful for (forward or inverse) kinematics engines that want to pursue fine-grained collision control with hands, fingers, body parts, etc.
3) Why viewpoints are necessary?
Again optional. Useful for (a) looking at body sections, for user navigation or medical applications, or (b) looking out from body locations, in a way that moves with any animation. For example, a set of viewpoints could provide multiple POVs for how a hand is interacting with a external object.
4) Are scale, scaleOrientation, and translation necessary inside Humanoid?
Again use is optional. I think they are useful for translation of different models, and also as possible animation effects. Further these fields provide HAnimHumanoid consistency with Transform, as do HAnimJoint and HAnimSite. Thus modeling, animation and implementation techniques can all be used consistently.
--- MWL: stated her questions for the WG to discuss
JW: How to treat ... as a skin or displacer technique
MWL: Why is the exact coordinates necessary for this example?
JW: I'm assuming we know what the mesh is. And we identify the boundary
MWL: We do not specify the boundaries you see in the snapshot. The boundaries are moving in animation so if we have one region with vertex (set) then they are moving.
We know the mesh, their order in the mesh and how the displacers will move them.
JW: The skin coordinate normals are optional but provide the facility for special effects in your animation. See examples in Don's book, X3D for Web Authors
DB: Any X3D browser can compute the size of Boxman's hands and use this information for collisions.
JW: If you leave out a bounding box the browser figures it out so you can safely ignore it. The way to figure it out is to experiment with it.
JW: Skin Normals allowed for a shading value when you had bones in humanoid. How the lights reflected and so forth. Circa with Keith Victor was active in the H-Anim WG.
JW: Viewpoints allows you to insert viewpoints with respect to humanoid so they are keyed when the humanoid moves.
JW: Try inserting a viewpoint nodes in you humanoid's face and see what renders.
DB: Scale, ..., Orientation? For consistency. Example, if someone says "I don't know how to use that or why that is there in the scene graph". There is an opportunity for animation I don't understand. How can I use that and innovate by applying old things in new ways.
DB: I think the BVH conversion algorithm should go in an Annex. If we can show a good mapping that covers most everything then that changes the impression of H-Anim from being a special approach to a general approach.
DB: five or six weeks ago Don posted a summary of the BVH work with Joe and MWL. Joe looked up in Blender; it imports in BVH then convert it to Quaternions! We have to incorporate quaternions and move away from Axis-Angle. Don asked, why do you have add that if we have bi-directional mapping? Joe said, Blender will try to correct the order if it is not "vYXZ"
Closed the meeting with next meeting date in April 2016 and topics, which will in the material William Glascoe emailed today (i.e., biometrics of the face and animating emotions and feelings)